With fears of mass unemployment and even societal destruction, we’re currently in an AI panic. But scientists and popular culture have been discussing the impact and risks of artificial intelligence for decades. Back in 1964, Fortune’s Gilbert Burck spoke with prominent computer scientists to get their predictions on what AI would be capable of in the years to come. So we took a look back at that 60-year-old article to see what they got right and wrong.
This piece was originally reported on by Jeremy Kahn, Fortune’s AI editor and author of Mastering AI: A Survival Guide to Our Superpowered Future.
This piece was originally reported on by Jeremy Kahn, Fortune’s AI editor and author of Mastering AI: A Survival Guide to Our Superpowered Future.
Category
🤖
TechTranscript
00:00For the last several years, it's felt impossible to avoid hearing about AI.
00:08Sometimes it's utopian promises of how it's going to make all of our lives easier.
00:11I think it'd be good to end poverty.
00:13Maybe we should stop a technology that can do that.
00:15I personally don't.
00:16Other times, it's fear of lost jobs, the spread of disinformation, and even apocalyptic scenarios
00:22brought about by these all-knowing machines.
00:24The creation of superhuman machine intelligence is probably the greatest threat to the continued
00:30existence of humanity.
00:31It has the potential of a civilizational destruction.
00:35But the last few years of rapid advancement in AI technology aren't the only time in history
00:39that computer intelligence has kept for attention.
00:41I'm sorry, Dave.
00:42I'm afraid I can't do that.
00:43Back in the 1950s and 60s, improving computer technology stoked fears that sound really
00:49similar to what we hear today.
00:51Sixty years ago, as computers learned how to write music and even win games of checkers,
00:55what did people think that artificial intelligence would be capable of?
00:58We dived into Fortune's archives to find out.
01:08Let's go back to 1964.
01:12The moon landing is still five years away and computers fill up entire rooms.
01:17But the technology and its capabilities have steadily evolved since the 1940s.
01:21The term artificial intelligence was created by John McCarthy, a computer scientist and
01:25professor at Dartmouth College who organized a now historic workshop on the topic in 1956.
01:32And by the mid-1960s, computers were capable of playing games of checkers, making music,
01:37answering simple questions about baseball teams, naming shapes, and recognizing voices.
01:42In the midst of all this progress, the public's perception of artificial intelligence drove
01:46fears of mass unemployment and concerns over computers' ability to produce art and make
01:50decisions more efficiently than the humans that built them.
01:53And while this conversation played out in American discourse, Fortune's Gilbert Burke
01:56spoke with prominent computer scientists and researchers to get an inside look at their
02:00predictions on how AI would impact the future, sharing his findings in 1964 with one simple
02:06question.
02:07Will the computer outwit man?
02:16Throughout his conversations, Burke heard a wide array of opinions concerning what computers
02:20would be able to do in the following decades.
02:22And some were very optimistic about how quickly artificial intelligence would reach major
02:26advancements.
02:27For example, Herbert Simon, a researcher at the Carnegie Institute of Technology, predicted
02:32back in 1957 that within 10 years, a computer would be crowned the world's chess champion
02:37that it would discover an important new mathematical theorem, and that it would write music of
02:41aesthetic value.
02:43Contrary to these predictions, it took exactly 40 years, not 10, for a computer to beat a
02:48chess champion.
02:49And a computer was first used to help prove the mathematical theorem in 1976, nearly 20
02:54years after the prediction.
02:55And although the first music composed by a computer was scored a year earlier in 1956,
03:00Burke stated that one of the moderate remarks about it is that repeated hearings tend to
03:04induce exasperation.
03:09Another expert with very hopeful predictions for artificial intelligence was Marvin Minsky,
03:13an influential early AI researcher who worked at MIT.
03:17Burke said that Minsky believed we were close to the threshold of an era that will quite
03:21possibly be dominated by intelligent problem-solving machines.
03:25Minsky also said that in 10 years, we may have something with which we can carry on
03:29a reasonable conversation.
03:30He even thought that if scientists work hard enough, they may have it in five.
03:34Well, Minsky turned out to have underestimated just how hard his colleagues were working.
03:39Just two years later, in 1966, ELIZA, the first computer chatbot, was unveiled by a
03:44fellow MIT computer scientist, Joseph Weizenbaum.
03:48ELIZA's programming was modeled after a form of psychology called Rogerian psychotherapy,
03:53meaning that it would rephrase what it was told into a question as a response.
03:57Well, my boyfriend made me come here.
03:59Your boyfriend made you come here?
04:01He says I'm depressed much of the time.
04:04I am sorry to hear that you are depressed.
04:06Weizenbaum's original intention was to demonstrate the shallow limits of a computer's ability
04:10to hold a conversation by essentially just having it repeat words back to users.
04:15But in a twist fit for a science fiction novel, users lost themselves in conversation with
04:20ELIZA, sharing very private thoughts with the one-dimensional computer, and resulting
04:23in what we now call the ELIZA effect.
04:26As a result, Weizenbaum changed his views on artificial intelligence and dedicated the
04:30rest of his career to criticizing the continued adoption of computer technology.
04:40But getting back to his article, Burke also heard from other experts who were a little
04:43more moderate in their predictions for AI.
04:46One was Arthur Samuel, a consultant to IBM's director of research and the originator of
04:50the term machine learning.
04:52In the 1950s, Samuel had successfully programmed a computer to learn how to play checkers and
04:57published his findings in 1959.
05:00But even after these accomplishments, Samuel was still a little skeptical about the computer's
05:04ability to surpass the human brain.
05:06Burke wrote,
05:07The limitations of the computer, Samuel likes to put it, are not in the machine but in man.
05:12To make machines that appear to be smarter than man, man himself must be smarter than
05:17the machine.
05:18Researchers had been attempting to build computers as complex as the human brain for years at
05:21this point.
05:22Burke noted that some experts he talked to thought the task would be way too expensive,
05:26saying,
05:27The total cost of duplicating all the brain cells and connections would come to more than
05:30one quintillion dollars, or one billion billion dollars.
05:35And that's in 1964 dollars.
05:37But despite the fact that computers are a fraction of the cost they were back in 1964,
05:41which has allowed computer technology to increase dramatically as a result, open AI is still
05:46looking for more computing power and a lot of money to continue improving their AI products.
05:52And Burke even shared his own thoughts about a computer's ability to generate new ideas,
05:55similar to a human brain, stating,
05:57Nobody has yet been able to program the machine to imitate what many competent judges would
06:01call true creativity.
06:03The computer's achievements in creative composition, literary and musical, are remarkable in the
06:07sense that Dr. Johnson's dog could walk on his hind legs was remarkable.
06:11It has not done well, but you are surprised to find it done at all.
06:16Coming back to the current day, it's clear that AI has evolved in a much different way
06:25than some of these computer scientists expected.
06:28Computers are so advanced that they're driving cars around cities and delivering takeout
06:31to hungry college students.
06:33Many of those optimistic expectations from Herbert Simon, along with those of his colleague
06:37Albert Newell, ended up coming true over the last 60 years.
06:41Their thoughts on a computer's ability to generate art are especially insightful, with
06:45the rapid advancement in text, music, image, and video generators we've seen since the
06:50late 2010s.
06:51Burke cited the duo as claiming,
06:53The computer on its own could not just copy but match such creations as a Beethoven symphony,
06:58crime and punishment, or a Cezanne landscape.
07:01It's true.
07:02AI technology is not only capable of writing books and making music.
07:06The technology can now make photorealistic images and even video.
07:10But something is still a little bit... off.
07:14Regardless, as the public perception of artificial intelligence shifts, and the economic threats
07:19and ethical concerns surrounding the tech seem to become more and more real, we're left
07:23with Burke's somewhat foreboding words from 1964,
07:27The computer is here to stay.
07:29It cannot be shelved any more than the telescope or the steam engine could have been shelved.
07:34Precisely because man is so arduously trying to imitate the behavior of human beings in
07:38the computer, he is bound to improve enormously his understanding of both himself and the
07:43machine.